VOL. 51, NO. 9
Journal of The American Ceramic Society
SEPTEMBER 21, 1968
Reaction Between Nitrogen and
Spinel in Chromium
D. M . S C K U G G S , * L. H. V A N V L A C K , a n d W. M. S P U K G E O N
I3cndix Rcsc;tt-ch Laboratories, Southfield, Michigan 48075, ; ~ n d
1)cp:uiiiciic oL Clicriiical and Metallurgical Enginccring, University of Michigaii, Ann Arbor, Micliigaii
The reaction between nitrogen dissolved in chromium and
a dispersion of magnesium aluminate particles in chromium a t high temperatures is discussed. The equilibrium state for various nitrogen concentrations in a 94
chromium4 spinel mixture was investigated between
1200° and 1400°C. Experimental techniques included
electron-beam microprobe analysis and mass spectrometry. Partitioning of nitrogen between chromium and
spinel was strongly in favor of the spinel. Massspectrometry indicated coordination of nitrogen with magnesium in the spinel lattice. The spinel-nitrogen reaction is
important because it can prevent the embrittling of
chromium by nitrogen.
481.04
r%>
CAl
-
MOLE
CAI
o KLEIN AND CLAUtR.nH = I9,000 MOLL
1
i l o i ~
to
06
I. Introduction
between metals and ceramics is assuming
greater impctrtancc as the demands of our technology increase. Strengthening ol metals with ccramic dispersoids
or with fibers and whiskers has received considerable attention. Modification of alloys by ceramic additions has
also becn attempted, primarily to increase hardness or heat
resistance.
Thcse modifications have proved very useful despite the
lowering of ductility in the composite which usually occurs.
This investigation concerns an exception to this effect on
ductility; that is, the considerable decrease in the brittleductile transition temperature of commercially pure chromium
by tlie addition of a spinel or spinel-forming oxide. In
gcneral, ductility as ineasurcd by elongation is reduced at
m y given temperature when a ceramic is added to a metal,
but when spinel is added to chromium a ductilizing effect
morc than offsets this normal embrittlement because the
brittle-ductile transition temperature is reduced.
This spinel effect is believed to be unique. It was first
described' in the system Cr-MgO-Cr,O,. A powder metallurgy preparation of 94Cr-6MgO had a brittle-ductile transition below room temperature with from 20 to 24% elongation
a L room temperature, which represented a drop in transition
temperature of nearly 400OC. X-ray examination of phases
cxtracted from the composite showed that considerable
quantities of magnesium chromite spinel had formed during
siiitcring by reaction of CrlOn in the powder with MgO.
On the basis of this work, the hypothesis of the present work
was that spinels absorb nitrogen dissolved in chromium and
Lhus eKectively remove it from the matrix, rendering the
matrix ductile.
Nitrogen is presently thought to be the chiei cause of the
brittle behavior of chromium at low temperatures.2-x Explanations of the embrittling phenomena vary, however:
some investigators favor Cottrell locking2s4; others, straininduced precipitation$; and some favor the action of nitride
l)rccipitatcs.G-8 The extremely low solubility of nitrogen in
chromium (Pig. 1) contributcs to thc uncertainty, as it is
tliliicult to make valid chcmical analyses atid to obscrve the
various tncchanisiiis. As little as 3 ppni of nitrogen has been
rcportcd lo cmbrittle clironiium.
07
Fig. 1. Solubility of nitrogen in chromium according to the
van't Hoff equation (Ref. 3).
I
INTERACTION
M t l A l IONS IN
1-
Fig. 2. Normal spinel lattice.
The prototype spinel, magnesium aluminate, was selected
as the oxide of primary interest because of its stability and
because it contains no chromium. Magnesium chromite was
also studied briefly.
The following description of the spinel structure was taken
from a description by Blasse.'"
The unit cell of normal spinel structure is shown in Fig. 2.
In the ideal crystal the anions form a close-packed cubic
Presented at the Fall Meeting of the Ceramic-Metal Systems
Division of The American Ceramic Society, French Lick, Indiana,
September 13, 1967 (Paper No. 19-C-67F). Received December
11, 1967; revised copy received April 26, 1967.
Based in part on a thesis submitted by D. M. Scruggs for the
Doctor of Philosophy degree in engineering materials at the
University of Michigan.
At the time this work was done, the writers were, respectively,
sctiior engineer, Bendix Research Laboratories; professor,
Department of Clicmical and Metallurgical Engineering, University of Michigan; and head, Materials and Processes Departiiicnt, Bendix IZcscarch LdboKttories.
* Now assistant professor, Departnicnt of Mechanical Etigjnccring, University of Arkaiisas, Fayettcvillc, kkdllSdS 72701.
473
fouuncd oJ' 7'he Americcm Ceramic Society- Scruggs el (1.1.
-1 7-1
Vol. 51, No.
1)
Table I. Raw Material Impurities
Cliroinium
Magiivsiuiii aluiiiiiiatc
0 . 009
. 003
0 . 001
0.001
,0025
,0054
0 . 56
. 60
(1)
where 0 is a cation vacancy. For every aluniinum atom
added L o lill a vacancy in formula ( I ) , charge compensation is
ncliicvcd by replacing three oxygcn ions with nitrogen ions.
'I'hus, AI&N is obtained by adding onc third of an aluminum
to formula ( I ) and compcnsating with the substitution of one
nitrogen lor onc oxygcn:
A15) iaO:iN
(2)
or
hl:~OjN
0,019
,0185
0 . 037
,014
0.003
striicture i i i which the cations occupy a fraction ol the tetra1ic.dral and octahedral interstices. Thcre are 32 anion posilions in thc uiiit cell, resulting in 63 tctraliedral and 32 octaliedral positions. Eight tetrahedral sites (callcd A sites)
are occupied by divalent ions, and 1 (i octahedral sitcs (callcd
U sitcs) arc occupied by trivalent ions. A nonideal structure is dcrived from the ideal one by moving the anions in a
11 I I ] direction away from the ncarest tetrahedral ion.
According t o ligand-field thewy, the octahedral site prefercnce for Cr is stabilized by i i h o ~ t40 kcal/mole.10 For
other ions, the site prefcrencc is less than half this energy.
Consequctitly, othcr spinels fcature trivalent ions in both A
Hlasse listed 147 binary spinels recorded
Nitrogen occurred in these spinels only as
,
the cyanide ion.
.4dams et d.''earlier reported a spinel in the system
9I2O1-A1N,corrcsponding to the lorniula A130aN. They wrote
~-A~,O
asJa tlclect spinel:
AMJi/aO.r
0.018
.016
( 3)
l'hcse writers also dol)ed AlaO:,N spinel with chromium and
concluded that Cr3+ would substitute for A1 in octahedral
sites only.
Tlic effects ol nonstoichionietry on possible reactions are
of considcrablc interest. Anderson, l 2 in a survey of the conseqttcnces of nonstoicliioinetry on the behavior of refractory
materials, statcd that in the distortcd spincl A&04 the coordination structures are based on the close packing of oxygens, and the encrgy of the substitutional defect pair is detcrniined principally by cation size. Furthermore, this
cncrgy is usually small, and all such compounds are likely to
undergo this limited typc of ordcr-disordcr transition at high
tcmpcratures.
Anderson referred to the work of Altman and Searcy,ls
who round that thc defect spinel Mg4A1(8-z2,a04
is fornicd above
I8OO0K, and to an investigation by k h a n of the reduction
of CaO with niclallic calcium. Only in tlic presencc of calciuin val'or i m 1 nitrogcn was a highly colorctl product lormcd.
' l l i v inatcrial coiitairictl Croin 0.85 to I .03 Ca,"
Incorporatioii of nilrogcn anioiints virtually to the creation of anion
vnrancics. Anderson statcd further that exccss electrons
may bc trapped as P centers, as altervalent cations, or as a
defect pair of zero virtual chargc. The close correlation between excess calciuni and nitrogen ions suggests that defect
pairs wcrc only slightly dissociated at 1 150° to .I20OoC and
that a n excess of calcium was nceded to forcc nitrogen into
tlic csscnlially ionic structure.
11. Experimeiital Procedure
M :t1 cLi%tls aiit l i)roccdLII-eswere sclcctccl to tlc Imiuiric tlic
iiiccli~~~iis~n
o f tluclility cnhancctneiit. IS(lui1ibriuiii sludies
\vci-c cloiie lo clctcriiiinc the amounts of constitucrits involved
0.15
0.50
Supplier
Ledoux
Supplicr
CHROMIUM
Fig. 3.
System Cr-spinel-N. Arrows indicate component
combinations investigated.
and analytical tcchniyues were used to determine nitrogen
content.
( 1 ) Materials
Ground electrolytic chromium* which analyzed !19.2Uj,
pure with oxygen as the principal impurity was used. The
powder was - 325 mesh (Tyler sieve), peaking at a particle
size of 10 to 1 2 ~ . Magnesium aluminate? was prepared by
fusion of stoichiometric amounts of high purity magnesium
oxide and aluminum oxide. This material was a -200 mesh
powder (Tyler sieve). Magnesium oxide and magnesium
chroniite were also used for control purposes. Chemical
analyses of the raw materials are given in Table I. Gases$
used were prepurified grades of anhydrous ammonia and
nitrogen and standard grade argon and oxygen.
Monochromiuni nitride was prepared by passing ammonia
over copper turnings a t 700OC and then over the chromium
powder at the same temperature for 24 hr. X-ray diffraction
ol the resulting powder gave a pattern identical to that listed
in the ASTM index for monochromium nitride. Dichromium
nitride was made by the same procedure, except that the
temperature was 1300OC. X-ray analysis demonstrated
that the resulting powder was pure dichromium nitride.
( 2 ) Experimental Method
Figure 3 shows the COllibindtiOnS of components €or which
the possibility of reaction was investigated. Significant combinations include chromium-nitrogen, which has already been
studied by Seybolt and OrianiI4 and by Caplin et aZ.15; spineliiionochroniium nitride; spinel-dichroinium nitride; and
spinel-nitrogen. Assuming that dissociation o€ the spinel is
insignificant, the system can be considered ternary. The
combination of most concern includes all three components;
the binaries are of interest only for elucidating the mechanism
oi the reaction. Procedures for the phase equilibrium studies
were :
(I) Determination of nitrogen reaction with spincls and
magnesium oxide, requiring furnace trcatinent of the oxides
in molecular and nascent nitrogen atiiiosphcres above 1000OC.
* Elcliroiiic-1,(2, Iliiioii
Carbitlc Corporatioii, New York,
N c w York.
1 Muscle Slionls l3lectruchciiiicnl
Coiiipaiiy,
'Tuscuiiibia,
Alnbarna.
$ Matheson Corporation, East Rutherlord, New Jersey.
Reaction Between Nitrogen and Spinel in Chromiuwt
475
T a b l e 11. Weight Changes D u r i n g Exposure of
(11) Study of the equilibrium bctween chromium nitrides
and spincl, using rurnacc trcatment of powder mixtures of
MgO, MgA1204, a n d M g C r 2 0 4to N i t r o g e n
September I!?(%
tlic tiiatcrials from 500' to 1445'C in dissociated ammonia.
( 3 ) Examination of the equilibrium compositions and
structures in the system chromium, spinel, and nitrogen,
which involved exposure of samples of wrought chromiumspincl composites to nitrogen a t low temperatures, equilibration a t I 'LOOo, 1300°, and 140Ooc, and subsequent analysis.
For the nitrogen reaction with spinel, fused silica boats
containing a I5 g powder sample were hcated in an 80% argon20% oxyg-en mixture for 2 hr a t the temperatures listed in
Table 11. The samples were cooled and then weighed on an
analvtical balancc. The boats were then heated in purified
ni-irogen for 3 hr a t the same temperature and reweighed.
The samples to be exposed to nascent nitrogen were heatcd in
anhydrous ammonia gas which had passed through copper
turnings heated to 900OC. The gas in the furnace was
checked with the ammonia quality apparatus described by
Ihillens'"; the dissociation was more than 95% complete.
Reaction o€ chromium nitrides with spinels was similarly
studied, using samples blended and spread evenly in the
I)ottom of ruscd silica boats held in an ammonia atmospherc
a t various tempcratures for 24 and 100 hr. The monochromiuni nitridc-spinel mixtures were heated at temperatures Irom
500' to 1000°C, and dichromium nitride-spinel mixtures lrom
1200° to 1500'C. Exact lattice parameters of the spinel
and o f the nitrides were determined by X-ray diffraction.
Composite specimens used in studying the chromium-spinelnitrogen system were prepared by powder metallurgy methods, i.c. blending of chromium and spinel powders, pressing
a t 15,000 psi, sintcring in palladium-purified hydrogen for 4
hr at 1450°C, and cxtruding at 1200°C with a 10: 1 area rctluction. Bar sainplcs wcre then equilibrated with varying
clitantitics of nitrogen.
A direcl method uscd by Seybolt and Oriani,I4 who detcrtniried ccluilibriuni partial pressures ol nitrogen in contact
with pure chromium a t saturation, was first attempted.
'They round the cquilibrium pressures a t 1 200°, 1300°, arid
1400'C to bc (j.(j*O.5, l O = k l , and 43.5AO.5 mm IIg, respectively. These pressures were maintained and regulated
by a controlled flow o€ argon and nitrogen, which were gettrred and introduced into a tube furnace. This method of
nitriding at low pressures was not entirely suitable, however,
hccause rathcr high temperatures were needed to achieve the
largcr nitrogen contcnts. Seybolt and Oriani pointed out
that chromium is an excellent oxygen getter a t 1200°C and
above and begins vaporizing a t 1300'C.
'4 capsule method of studying the system was therefore
uscd, eliminating thc problems of oxidation and vaporization.
I h - s of the coniposjtc weighing approximately 10 g were
nitridcd a t 1000' and 1 LOOC
' for various times. At these
tcmperatures the use of prepurified nitrogen, further purified
by passage over hot copper turnings as described by Seybolt
arid Oriani, resulted in a reproducible weight change in the
samplc. A slight tarnish was evident but no green oxide
rornied, and no evidcnce of chromium vaporization was noted.
These samples were then scaled in evacuated fused silica
capsules, heated lor 48 hr a t 1200", 1300°,and 14OO0C,and
quenched in ice-brine. No visible reaction occurred between
the samples and the silica, Phases in the resulting samples
wcre idcntified metallographically. A transverse section of
each specimen was broken off, cold-mounted, and polished.
The appearance of primary nitride (formed a t equilibrium
and present beforc quenching) and precipitated nitride represented the significant data points.
Kingcryl' discussed the difficulties of using phase disappearance in equilibrium studies, the principal one being that of
identifying a new phase by microscopy or by X-ray diffraction.
I-Ie statcd that as much as 10% of a new phase may be necessary for the X-ray method, buL as little as 0.5% for microscopy, An attempt t o lower the limit for microscopy was
madc in this study by using the electron microscope for com-
Temp.
("C)
Sample
1100
MgA1204
Time
(hr)
W t change
(g/g)
Molecular nitrogen
MgCrzO4
3
3
-0.007
4- ,001
,000
-
,000
.001
Dissociated ammonia
1200
MgA 120n
MgCrsOe
3
3
,000
1445
MgA12Ch
4
4
4
150 ppm Ns*
1500
MgCrsOr
MgCreO4
,000
170 ppIt1
300 pptll
-0.01
N2*
N%*
* By chemical :inalysis.
positions near the phasc boundary; the electron microscope was essential for distinguishing between small nitride
particles and etch pits. Specimens for electron microscopy
were prepared by two-stage plastic replication. Platinum
carbon shadowing a t a low angle (30') intensified the contrast of fine structurcs.
Analyses for chromium and other metallic elements were
performed by emission spectrography. Nitrogen analyses
were by the Kjeldahl method. Electron microanalysis was
used for selected samples to provide a qualitative in situ
determination of nitrogen. Electron-beam microprobe analyses were conducted with a tungsten emitter; a lithium
stwratc crystal, and a flow counter on standard AKL units
ISMX and AMX. Standards were based on the nitrogen
peak emitted from horon nitride and dichromium nitride.
Temperatures bclow 1200'C were measured with ChromelAluinel thermocouples cemented to the furnace tube wall
at the position of the sample. Temperatures above 1200OC
were measured with a W-W26Re thermocouple in hydrogen
and ammonia and with a Pt-Ptl3Rh thermocouple in air
and argon.
A Reichart Vacutherm hot-stage microscope was used to
study the microstructure from room temperature to 900°C
a t a vacuum of 5 X
torr.
A ncndix Time-of-Flight mass spectrometer, model 1111,
was used to identify mass numbers of gaseous prodvcts
emitted by the chromium-spinel compositc from room temperalurc to XOOOC in high vacuum.
111. Results
( 1 ) Spinel-Nitrogen
The weight changes which occurred on heating magnesium
oxide, magnesium aluminate, and magnesium chromite with
atomic and molecular nitrogen ate given in Table 11. The
data clearly indicate that no reaction tales place a t 1100O or
1200OC between. any of the oxides and nitrogen in either
form. At very high temperatures, some nitrogen absorption
occurs in the spinels.
The portion of the investigation concerned with the oxidenitrogen system thus indicates two specific results:
(1) Magnesium aluminate, magnesium chromite, and
magnesium oxide are chemically inert to atomic or molecular
nitrogen at 1100' and 1200°C.
(2) A small quantity of nitrogen reacts with spinels above
1400°C, resulting in a pickup of a few hundred parts per
million nitrogen.
( 2 ) SpineLChromium Nitrides
The chromium content of magnesium aluminate spinel
equilibrated with chromium nitrides, as determined by lattice
Journal o j The A rnerircin Cercrrnir Society-Scruggs
c.t ctl.
Vol. 51, No. !I
TOTAL NllROGEN CONTENT . P P M
TEMPERATURE -
"C
Fig. 4 . Chromium content of spinel heated with mixed
chromium nitrides (curve 1) and with chromium (curve 2 ) ,
as a function of temperature. No difference in chromium
content was noted between temperatures where CrN is more
stable, 700' to 800"C,and the temperature region where CrsN
is more stable, 1
to 1200°C.
Fig. 5. Acicular nitride phase in 94 Cr-6 spinel composite as
a function of nitrogen content and temperature. Pure Cr
without spinel exhibits acicular nitride at 1 5 to 2 5 ppm nitrogen level. Data points are coded with sample numbers.
paranictcr shift, is shown in curvc I , Fig. 4. X-ray dirfraction of the saniples aftcr equilibration indicated mixed CrpN
and CrN from 750° to 10I)O°Cand CrzN only above 1200°C.
Little rcactiori was evident up to l:lOO°C. Coloring of the
spinel (pink) confiriiicd the observations. Faint coloring
appeared at 1300OC and deepened considerably a1 1445°C.
This phase of the experiments indicated that chromium
readily dissolves in magnesium aluminate above I 300°C,
but no reliable chemical analysis could be made for nitrogen
because of the difliculty of separation. No new phase appeared
in thc sl'incl at any of the temperatures investigated.
( 3 ) Chrombm-SVine~-N~trogen
Exposure to nitrogen and subscqucnt analysis of bars of
wrought chromium containing ri% spinel (magnesium alumiuatc) was the last part of thc component study. Samples
made from pure chromium powder, without spinel, contained
about 50 ppm nitrogen and had quantities of precipitated
nitrides in the extruded condition, in agreement with previous work.
Samples of chromium containing 6 wt% spinel behaved
quite diflerently. No nitrides were prcscnt after fabrication
and extrusion. The samples were then exposed to varying
c~uantitiesof nitrogen a t I O O O O and 1100°C. When the nitrogcw charge exceedcd the solubility limit, a riitridc case was
forined. 'l%is case disappeared, however, when the samples
wcre held a t tempcrature. Microstructures, aftcr heating
for I8 hr, were rrce of nitrides and were equivalent to slructures of samplcs heated for 48 lir. Turkdogan arid Ignatowicz'* reported that equilibrium was reached after 48 to 52
hr above llOO°C in studies of nitrogen ecjuilibraCion with ironcliromium alloys.
T o ensure equilibrium, all the samples were held for 48
lir a t 1200°, 1300°, and 1400°C. As the amount of nitrogen
in the samples increased, the chromium was first completely
free of nitrides or any other feature; more nitrogen caused
pitting after etching with a 3 : 1 glycerine-HC1 solution.
Quantitics of etch pits appeared, and finally, needles of
nitride were seen as excess nitrogen caused well-defined
primary nitride crystzals to form a t the equilibration temperature. These were first noted in the grain boundaries.
X-ray examination of spinels extracted with brominemethanol lrom three control samples which were unnitrided
hut equilibrated a t 1200°, 1300°, and 14OO"C, respectively,
demonstrated that spinel in the chromium matrix absorbed
more chromium at 12OOO and 1300°C than when heated with
tlichromium nitride. After cooling, many spinel grains
contained a precipitated chromium phase. The chromium
content computed by lattice parameter shift is shown in curve
2, Fig. 4.
The heat-treated 94 chromium-6 spinel composites can be
divided into two classes: those which contain acicular nitridc
Fig. 6. Sample 16, with 140 ppm nitrogen, equilibrated at
1200OC. Very little pitting and no nitride needles. Dark
areas are spinel of widely varying grain size. HCl-glycerine
etch ( X 9 0 0 ) .
Fig. 7. Sample 17, with 180 ppm nitrogen, equilibrated at
1200'C. Very extensive etch pitting, as shown in circled area,
but no nitride needles. HCI-glycerine etch ( ~ 9 0 0 ) .
precipitates, and those which do not. Pure electrolytic
chromium control samples (without spinel) similarly processed but not exposed to nitrogen exhibiled gross nitride
precipitates after treatment a t all temperatures.
The composition points for extruded 94 chromium4 spinel
are plotted in Fig. 5. A boundary line separates the compositions with and without acicular nitride precipitates at
room temperature. Electron microscopy was used to identify
September 1968
Fig.
8.
t400OC.
Reaction Between Nitrogen and Spinel in Chromium
Sample 21, with 7 3 0 ppm nitrogen, equilibrated at
A few widely separated nitride needles as shown in
circled area. HCl-glycerine etch ( X 9 0 0 ) .
477
Fig. 10. Sample 2 0 , with 4 2 0 ppm N, equilibrated at 1200OC.
Large chromium grains, many containing large nitride needles.
Circle encloses a chromium oxide particle. HCI-glycerine
etch ( X S O O ) .
Table 111. Electron-Beam Microprobe Determination
of Nitrogen in Sample 34 (Nitrogen Content 0.098y0)
Phase
Mean
(cts/sec)
Sample
size, n
Cr nitride, CrzN
Cr needles
Cr grains
Oxides, low N2 content
2014
246*
224 t
173
5
10
10
37
* Neglecting a reading of
485.
t Neglecting readings of 637, 684, and 380.
Table IV. Nitrogen X-Ray Emission in Selected
Spinel Crystals in Sample 34*
Fig. 9. Sample 3 4 , with 9800 ppm nitrogen, equilibrated at
1400°C. Contains Cr with nitride needles (areas marked c),
and Cr& with Cr needles (areas marked n). HC1-glycerine
etch (X500).
nitrides in the samples. Characteristic microstructures are
shown in Figs. 6 through 10.
( 4 ) Chemical AnaZysis
I n the Kjeldahl method, the chromium-spinel sample is
first digested with HCl. The spinel is not dissolved and can
be separated by filtration. Nitrogen contents of the separated spinels were extremely low and quite variable. The
solutions tested high in nitrogen and gave consistent
results. I n view of the microstructures, the only explanation
was that the portion of the spinel that contained the nitrogen
was dissolved by HCI. This result suggested the possibility
of microanalysis in situ with a n electron-beam microanalyzer.
( 5 ) Electron-Beam Microprobe Analysis
Recently developed techniques for analysis of the soft X
rays emitted by nitrogen were used. An electron-beam
microanalyzer was used to examine individual oxide crystals
at random in samplc 34 which was equilibrated at 14OO0C.
The nitrogen peak was scanned for maximum reading with
boron nitride. A chromium-spinel sample not exposed to
nitrogen and contahing 50 pprn total nitrogen was used for
the background count. Essentially the results are as would
be expected, except that the oxides in sample 34 could be
Toedl
reading
Counts/sec
attributed to
nitrogen
3 14
490
289
328
309
478
801
138
314
113
152
133
302
625
(%I
0.73
1.69
0.56
0.81
0.72
1.65
3.3
Average 1.3?0
N2
* These crystals have statistically significant nitrogen contents computed on the basis of 11%Nz in CrzN.
separated into two groups, one group apparently containing
no nitrogen. Results are given in Tables I11 and I V .
The random appearance of nitrogen from one oxide particle
to another posed serious questions of distribution and chemical
activity. The full composition of single oxide crystals in
situ was therefore investigated. It seemed highly possible
that A1N or Mg,N2 was present, so that careful sample preparation was necessary because of their property of rapid hydrolysis. Samples for metallography and microprobe analysis were cut and polished in absolute alcohol; no water or
etchants were used and very little heat was generated in
mounting and grinding.
Electron backscatter of an interesting area is shown in
Fig. 11. Several oxide particles of varying size and shape are
in the area. One of particular interest is centered in the
photograph and shows a concentration of heavy element
478
Journal of The American Ceramic Society-Scruggs
Fig. 1 1. Electron-beam microprobe photograph of electron
hackscatter of sample 43, area ( l ) , no etch ( X 1200). A line
trace made along the grid marked with arrows appears in
Fig. 15. Composition of the large oxide grain is apparently
somewhat different than that of the smaller oxides. Sample
was equilibrated at 1300°C and has a nitrogen content of 1360
PPm.
et al.
Vol. 51, No. 9
Fig. 13. Electron-beam microprobe photograph of chromium
X radiation from area (I), Fig. 11. No etch ( X 1200).
Fig. 14. Electron-beam microprobe photograph of magnesium
X radiation from area (I), Fig. 11. No etch ( X 1 2 0 0 ) .
Fig. 12. Electron-beam microprobe photograph of oxygen
X radiation from area ( I ) , Fig. 11. No etch ( X 1200).
(chromium) in the center. This area is coded area (1) and
Fig. 1% shows the Characteristic oxygen I< line from the
sample. Similarly, Fig. 13 is a photograph with chromium
radiation and Fig. 14 with magnesium. The chromium-rich
oxide is apparently low in magnesium. The small spinel
grain to the right of the center and the grain on the far right
have less chromium and more magnesium.
Figure 15 is the trace of measured radiation characteristic
of the magnesium, oxygen, chromium, and nitrogen in the
sample, lcft to right along the line marked by arrows in Fig.
11. Thc high magnesium content in the small inclusion seen
in the characteristic radiation photograph is apparent.
A second area of the same sample of Fig. 11 demonstrated
a disparity between magnesium and aluminum contents.
Again, where the magnesium content was elevated the chromium content was depressed.
A number of scans were performed in a third area, and
wherever magnesium content was significant a nitrogen reading several times background count was apparent. Longtime counting was attempted, but the nitrogen count rate
decreased with time, presumably because of heating by the
electron beam.
A fourth area was examined. The characteristic magnesium radiation was photographed, and a trace was made
across the specimen through two spinel particles, one having a
high magnesium content. The beam traces for magnesium,
chromium, and nitrogen are shown in Fig. 16. The nitrogen
count in the right hand spinel particle is statistically significant, representing 0.75% nitrogen.
(6)
Hot-Stuge Microscopy
The apparent decomposition in the electron beam of the
spinel phase, which should be quite stable, suggested the
use of a vacuum hot-stage microscope to observe the reaction
directly. Equilibrium sample 44, made with reagent purity
spinel with approximately 3000 ppm nitrogen added, was
Reaction Between Nitrogen and Spinel in Chromium
September 1968
479
E
w
I
I
c
I
10
30
20
MASS NUMBER
Fig. 17. Mass spectral intensity of gases from sample 4 3
heated to 75OoC, as a function of mass number. This sample,
containing 1360 ppm nitrogen, exhibited a mass number peak
at 3 8 , corresponding to MgN. Other peaks are background
corresponding to water vapor, 0, and Nz, and are present at
all temperatures. Note that the abscissa is not a linear scale.
Each square represents 17p.
DlSlANCE
IV. Discussion
- MICRONS
Fig. 1 5 . Line trace of intensity of characteristic X rays from
sample 4 3 , area (I), Fig. 11. No significant difference in
nitrogen level is apparent. The small inclusion has more
than twice the magnesium content of the large inclusion.
Mg 11,000
n
I1
---- MAGNESIUM
CHROMIUM
NITROGEN
~
II
,
I
I,
I
I1
Aj
-31*
-25,000 CI
-140 N
-
Mg 116
~
68
DISTANCE
136
204
- MICRONS
Pig. 16. Line trace of intensity of characteristic X rays from
sample 4 3 , area ( 4 ) . The nitrogen level in the right hand
oxide particle is significantly above background and indicates
that nitrogen is associated with the higher magnesium content.
heatcd in the Vacuthcrm to 700°C a t
torr Angular,
highly colored spinel grains were unaffected by the heat,
whereas the clear white or light yellow grains were definitely
affected Surface roughening, similar to thermal etching,
appeared randomly throughout each of the oxide grains,
occurring around the edge, across the center, or throughout
the particle
( 7 ) Muss Spectrometer Examination
A small amount of sample 43 (94 chromium-6 spinel expoied to nitrogen) was prepared by dry-grinding and heated
in the ionization chamber of a Bendix Time-of-Flight mass
spectrotncter No significant peaks above background appeared until 750°C was reached, when a very strong peak
occurred a t mass number 38 (Fig 17)
This peak could only be caused by a molecular combination
of Mg and N as MgN No other known combination of the
matcrials present (Cr, Mg, Al, 0 , N) would yield this mass
number After a few minutes a t 75O"C, further increases in
temIierature qave no indications above background
Reactions of Spinel with Nitrogen and Chromium
Nitrzdes
(1)
Very little reaction was evident between magnesium aluminate or magnesium chromite and either molecular or nascent
nitrogen. The small amount of nitrogen absorbed at 1500°C
could be explained by inaccurate analysis caused by breakdown of spinel containing nitrogen in the moisture present in
ordinary air. This result, if real, could also be explained by
a small favorable free energy difference in the substitution of
nitrogen for oxygen in the spinel and reaction of displaced
oxygen with the hydrogen available from the dissociated
ammonia.
The extent of reaction of magnesium aluminate with either
chromium nitride could not be studied by analyzing for nitrogen content. The extent of reaction of the spinel with chromium, however, could be adequately determined. It is evident
that the chromium nitride must break down before significant
solution of nitrogen takes place. The larger concentration
of chromium dissolved in spinel dispersed in the metal (curve
2, Fig. 4) can be contrasted with the chromium concentration in spinel exposed to the pure nitride (curve 1, Fig. 4).
Dissociation of CrN a t 1000°C and below did not affect the
concentration of chromium in the spinel. The most interesting case is that of spinel in the chromium matrix containing
nitrogen in solution. This is discussed in the next section.
(2) Reaction of Chromium-SpineLNitrogen
Compositions
Small amounts of nitrogen in chromium precipitate $t low
temperature because of the limited solubility of nitrogen in
chromium. Caplan et u Z . ' ~ published a photomicrograph of
chromium containing 60 ppm nitrogen which exhibited extensive Widmanstatten precipitate of dichromium nitride.
This type of precipitate was noted in all the spinel-free chromium samples used in this study, as well as in three samples
which contained magnesium chromite spinel. Weaver5 noted
precipitates a t the 50 ppm level of nitrogen, in agreement
with this study. The minimum nitrogen level causing precipitation is thus extremely low. In the results plotted in Fig. 5
the boundary line is estimated to be accurate to within 20 ppm.
The microstructures of Cr-spinel composites which did not
contain nitrides prove that nitrogen in solution is partitioned
between the chromium and the spinel, and that solution in the
spinel is the most Favored case. For example, sample 37
has a nitrogen content of 650 ppm overall. The matrix is
free of nitrides and contains no more than 20 ppm nitrogen.
This means that the spinel at 6 wt% contains about 1% nitrogen, in agreement with electron-beam microanalysis. There
apparently is also a very slow increase in the nitrogen content
in the matrix, as shown by the etch pits exhibited by sample 18,
equilibrated a t 1200"C, and sample 36, equilibrated a t
130OoC, after quenching. These pits are apparently caused
by nitrogen clustered around dislocations. This structure
4SO
Journal of The American Ceramic Society-Scruggs et al.
evidently results whcn there is enough dissolved nitrogen
(probably less than 5 to 10 ppm) to cause clustering but not
enough to form precipitates. The pits may be confused
with precipitates in optical micrographs but may be seen
clcarly in the electron microscope.
The microstructures indicate that nitrogen is absorbed by
spinel and that the partition of nitrogen between chromium
and spinel is strongly weighted in favor of the spinel a t low
nitrogen levels.
( 3 ) Chemical Analysis and Microprobe Analysis
Chemical analysis of the composites does not separate
nitrogen in spinel from nitrogen in chromium. Adams et aLl1
reported that A1N in the system A1203-A1Ncould be dissolved
from an Al3O3Nspinel structure with dilute NaOH.
Thc breakdown of many of the spinel grains by heating in
the electron beam and by etching suggests that the nitrogen
is covalently bonded and probably tetrahedrally coordinated.
Hlassel’ stated that the controlling situation for cation distribution in spinel is that electrostatic considerations lead to
very small energy differences for ions other than chromium,
and if bonds are stabilized by an effect of nonelectrostatic
origin such as covalent bonding, that type of distribution will
probably have the lowest energy.
This was confirmed by the mass spectrometer experiment.
The nitrogen apparently coordinates with a magnesium ion
in the spinel, thus becoming trapped in the spinel lattice and
hcing effectively removed from the chromium matrix. When
exposed to the environment at the surface, the nitrogenspinel combination acts very similarly to free MgN. The
‘i50”C decomposition temperature in the m.ass spectrometer
corresponds to 700°C for the dissociation observed in the
hot-stage microscope. These results can be compared to
those of Brewer et ul.,19 who reported that trimagnesium
dinitride (Mg8N2)decomposes at 113OoC.
The electron-beam microprobe analysis clearly demonstrates that many of the oxide grains have statistically significant nitrogen contents. The fact that many particles do not
rcgister nitrogen may well b-, a function of sample deterioration due to moisture and heat sensitivity. The resulting
surface roughness can also significantly lowet the X-ray count.
The nitrogen also appears to coexist with an enhanced rnagnesium content, confirming the coordination of magnesium
with nitrogen.
At least two important conclusions can be drawn from the
chemical analysis of the chromium-nitrogen samples. Complete acid digestion of CrN is difficult, or impossible. Formation of aluminum or magnesium nitride, even combined in a
spinel phase, requires the same careful handling that the
nitride alone would demand. The last conclusion also has
relevance in the study of aluminum nitride in nitrided steel,
i.e. the AlN may be completely removed by wet grinding or
heating during prcparation of samples for metallographic
cxamination.
( 4 ) Mechanism of Nitrogen Removal
The plausible mechanisms of removal are :
( A ) Interstitial Solution of Nitrogen in Spinel: This
would be characterized by decreasing solubility with decreasing temperature. The normally low free energies of
solution of this type would indicate either loss of nitrogen at
low temperature or precipitation of a nitride a t low temperature.
( B ) Substitutional Solution of CrN and/or Cr2Nin Spkel:
This would be characterized by nearly stoichiometric substitution of the nitride in the spinel, wherein a continuous
change in lattice parameter of Lhe spinel would occur until
saturation of the spinel and the appearance of a new phase.
The cubic lattice structure of CrN suggests that such solution
would bc favored and should peak a t or below the dissociation
temperature of CrN if this mechanism occurs The ecluations for this reaction are:
xCrN
Vol. 51, No. 9
+ M~AIzOOI
F’ MgAlzCr,OaN,
xCrzN 4- MgAlnOa
* MgAlzCr,,OaNz
(41
(5)
(C) Reaction of Nitrogen with Spinel in the Presence of
Chromium: This would be characterized by dependence
on defect formation in the spinel, which would require that
chromium nitride be dissociated before the reaction could
proceed.
Step 1:
Step 2:
CrzN F? 2Cr
25
4-&
+ 3MgAlz04 * 3MgAlaOaNm + 3 0
(6 )
(7)
No precipitation or other evidence of decreasing solubility
was found, and thus mechanism (A) is not likely. No evidence of direct solution of either CrN or Cr2N in spinel was
discovered (Fig. 4). Moreover, increased nitrogen content
was associated with increased magnesium content but not
with increased chromium content. Thus mechanism ( B )
does not seem reasonable. The most likely mechanism is
thus substitution (Eqs. (6) and (7)). We envision a continuous series of spinel structures in which chromium is substituted for aluminum and nitrogen for oxygen coordinated
with magnesium.
Andersonlz discussed a similar situation in regard to the
reaction of nitrogen and calcium with calcium oxide. He
stated, “Incorporation of nitrogen (in calcium oxide) virtually amounts to substitutional mixed crystal formation
with calcium nitride, creating a corresponding number of
anion vacancies.”
The corresponding situation, that of Eq. (7) above, appears
to fit all the data. Other arguments that favor this conclusion are as follows:
(1) Gaseous nitrogen has no effect on spinel in any form
to 1200°C and slight effect in the form of dissociated ammonia
at 1400°C and above where nonstoichiometry occurs, and
where hydrogen can be assumed to react with any released
oxygen.
(2) Chromium is found in considerable quantity in spinel
a t 1200°C and above.
(3) Chemical and microprobe analyses prove that chromium and nitrogen do not exist in spinel in any stoichiometric
ratio. Dissociation proceeds by separation of MgN.
(4) Microprobe analysis and hot-stage microscopy indicated a nitrogen spinel with properties similar to those of the
AI3O3N spinel reported by Adams et aLl1
( 5 ) Oxygen released by the substitution can react readily
with chromium in the matrix.
V. Conclusions
1, Partitioning of nitrogen between magnesium aluminate
spinel and the chromium matrix has been shown to occur.
The spinel contains 170 ppm of nitrogen a t 1200°, 385 ppm
a t 1300”, and 680 ppin at 1400”C, where the chromium is
saturated with nitrogen after cooling to room temperature.
2. The mechanism of nitrogen removal is one of formation
of N3- ions in anion vacancies present in spinels above 1200°C
and subsequent coordination with magnesium in the spinel,
in agreement with field theory.
3. Dissociation of the nitrogen-containing spinel proceeds
by evolution of MgN.
4. Chemical analysis of this system is not possible by phase
extraction since the nitrogen-spinel combination breaks down
in the extraction process.
5 . Electron-beam microprobe analysis for nitrogen in
spinel is possible on a semiquantitative basis if care is taken
not to decompose the nitrogen-spinel combination during
sample preparation, and if the decomposition under the electron beam is recognized.
Acknowledgments
The authors are indebted to the Bendix Corporation for
sponsorship of this work. The use of electron-beam microprobes
at the ARL Laboratory in Dearborn and at the Ford Motor
Company Research Center is also gratefully acknowledged.
September 1988
Crystallite Growth of Be0 Powders and Its Inhibition by Adsorbed Phosphate
References
1 D. M . Scruggs, “Modified Chromium for Unprotected
Structures,” A R S ( A m . Rocket Soc.) J . , 31 [ l l ] 1527-33 (1961).
F. Henderson, S. T. M. Johnstone, and H. L. Wain, “Effect
of Sitride-Formers upon the Ductile-Brittle Transition in
Chromium,” J . Inst. Metals, 92 [4]111-17 (1963-64).
3 M. J. Klein and A. H. Clauer, “Sitrogen-Induced Internal
Friction in Chromium,” Trans. A I M E , 233 [9] 1771-77 (1965).
F. P. Bullen, F. Henderson, H. L. Wain, and M. S. Patterson, “Effect of Hydrostatic Pressure on Brittleness in Chromiuin,” Phil. Mag., 9 [ l o l l 803-15 (1964).
C. W. Weaver, “Strain-Age Hardening and Brittleness in
Chromium,” Nature, 180,806-808 (1957).
A. Gilbert and M. J. Klein, “Effect of Cooling Rate on the
Ductile-Brittle Bend-Transitiou Temperature of Chromium
Wire,” A d a Met., 14 [4] 5 4 1 4 3 (1966).
A. Gilbert, C. S.Reid, and G. T. Hahn, “Observations on the
Fracture of Chromium,” J . Inst. Metals, 92 [ill 351-56 (196364).
K. E. Hook and A. M. Adair, “Recrystallization Embrittlement of Chromium,” Trans. A I M E , 227, 151-59 (1963).
D. J. Maykuth, W. D. Klopp, R. I. Jaffee, and H. B. Goodwin, “Metallurgical Evaluation of Iodide Chromium,’’ J . Electrochem. Soc., 102 [6] 316-31 (1955).
lo G. Blasse, “Crystal Chemistry and Some Magnetic Properties of Mixed Metal Oxides with Spinel Structure,” Philips Res.
Refits., S ~ f i p l .3,
, 1-139 (1964).
48 1
I. Adams, T. R. AuCoin, and G. A. Wolff, “Luminescence
in the System A120,-AlN,” J . Electrochem. SOL, 109 [11] 1050-54
(1962).
Ameril 2 J. S. Anderson; in Son-Stoichiometric Compounds.
1963.
can Chemical Society, Washington, D.
l 3 R. L. Altman and A. W. Searcy; in Symposium on Chemical
and Thermodynamic Properties at High Temperatures, 18th
International Congress of Pure and Applied Chemistry, Montreal,
Canada, 1961.
l4 A. U. Seybolt and R. A. Oriani, “Solubility of Nitrogen in
Chromium,” J . Metals, 8, 556 (1956).
l 5 D. Caplan, M. J. Fraser, and A. A. Burr; pp. 196-215 in
Ductile Chromium and Its Alloys. American Society of Metals,
Washington, D. c.,1957.
l6 D. K. Bullens, Steel and Its Heat Treatment, Vol. 11, 5th ed.
John Wiley & Sons, Inc., S e w York, 1948.
l7 W. D. Kingery, Property Measurements a t High Temperatures. The Technology Press of Massachusetts Institute of
Technology, Cambridge, Mass., and John Wiley & Sons, Inc.,
New York, 1959.
18 E. T. Turkdogan and S. Ignatowicz, “Solubility of Nitrogen
in Iron-Chromium Alloys,” J . Iron Steel Inst., 188 [3] 2 4 2 4 7
(1958).
l 9 L. Brewer, L. A. Bromley, P. W. Gilles, and N. L. Lofgren;
Paper No. 4 in Chemistry and Metallurgy of Miscellaneous
Materials; Thermodynamics. Edited by L. L. Quill. McGrawHill Book Co., New York, 1950.
l1
c.,
Crystallite Growth of Beryllium Oxide Powders
and Its Inhibition by Adsorbed Phosphate
G . H . P R I C E , W . I . S T U A R T , a n d D. G . W A L K E R
Australian Atomic Energy Commission Research Establishment, Lucas Heights, New South Wales, Australia
Low-temperature sintering of beryllium oxide powders
from 300° to 1000°C was studied by measuring crystallite growth and changes in specific surface. Sintering
is enhanced by the presence of water vapor. Adsorbed
phosphate almost completely inhibits sintering up to
850”C,even in water vapor at 10 torr.
could modify or inhibit sintering. Therefore, low-temperature sintering of beryllium oxide powders was investigated,
using surface area measurement, X-ray diffraction techniques,
and electron microscopy, with special emphasis on the adsorption of phosphate on beryllium oxide and its effect on sintering.
11. ExDerimental Procedure
I
I. Introduction
between about 300” and 1000°C of high-area
metal oxide powders often leads t o crystallite growth and
reduction in specific surface; t h a t is, some form of lowtemperature sintering occurs. Water vapor usually enhances
this t y p e of sintering. Anderson and Morgan’ showed t h a t
rates of crystal growth in magnesium oxide powder can be increased more t h a n lo3 times in t h e presence of water vapor,
and steam-sintering of silica and silica-alumina catalysts has
been studied.2p6 Rau7 demonstrated t h a t considerable
crystal growth occurred in beryllium oxide powders on heating
in air below 1000OC.
Mechanisms of low-temperature sintering are not clearly
understood, b u t de Boer and Vleeskens5 and Anderson and
Morgan‘ suggested t h a t enhancement in t h e presence of water
vapor occurs because of condensation between surface hydroxyl groups of adjacent crystallites with formation of
oxygen bridges. T h e possibility then arises t h a t , if hydroxyl
groups are essential intermediates during sintering in water
vapor, the prescncc of other stable inorganic surface groups
C
ALCINATION
Received Deccmbcr 18, 1967.
The writers are, respectively, experimental olIicer, scuior research scientist, and principal research scientist, Australian
Atomic Eiicrgy Commission.
(1)
Preparation of B e 0 Powders
Powders with surface areas of about 300 to 400 mz/g were
prepared by vacuum thermal decomposition of P-Be(OH), or
BeSO4.4IIZ0. Hydroxide-derived B e 0 (H-BeO) was prepared b y heating t o 28OoC, and sulfate-derived material
(S-BeO) b y heating t o 950°C. T h e temperature was raised
slowly t o maintain the pressure of gaseous decomposition
products below 0.2 torr. Isothermal calcination was then
continued until the pressure decreased t o about 0.02 torr.
T h e exact form of this temperature program depended on the
amount of material used and the geometry of the system.
Excellent batchwise reproducibility was obtained in this way.
T h e B e 0 powders were then washed with distilled water
until the conductivity of the wash water decreased t o about
0.5 pmho/cm2.
( 2 ) Meastlrement of Crystallite Size
Average crystallite sizes were calculated from X-ray line
broadening. X-ray diffraction profiles for the (100) and
(002) Cu Kcv reflections were obtained using a Philips diffractometer with a chart recorder. Integral breadths were
measured for the recorded profiles, corrections for instrumental broadening were made b y the method of Klug and
Alexander,8 and the crystallite size was calculated from the
Scherrer formula.